Release 1.2.0: learned confidence classifier, graph/vault benchmarks, transform tests#5
Merged
Conversation
… transform tests - Ship the secret-confidence classifier (PR #4) as 1.2.0's headline: opt-in refineConfidence scoring for generic detectors, exported via the new flare-redact/ml subpath and --refine-confidence in the CLI. - Add bench/graph-vault.mjs (npm run benchmark:graph): redact() on flat strings and deep object graphs, and vault mint+restore round trips. - Add direct unit tests for pseudonymize/surrogate (Luhn-valid card surrogates, keyed determinism) and mapGraph traversal (URL, URLSearchParams, Error prototypes, shared refs, sparse arrays, symbol keys, Map/Set). 162 -> 170 tests.
…e coverage - phone (opt-in) now matches formatted national numbers ((555) 123-4567, +90 532 123 45 67, trunk-0 forms) with digit-count validation and a date guard; bare digit runs still never match. - New default-on detectors: gcp_service_account (private_key_id in service-account JSON) and gcp_refresh_token (1// OAuth tokens). - CI now smoke-tests the dependency-free core on Bun and Deno via scripts/runtime-smoke.mjs; publish depends on both passing. - README: React/browser usage section. 170 -> 174 tests.
umudhasanli
approved these changes
Jul 24, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Ships the learned secret-confidence classifier (#4, thanks @Doofz!) as the headline of 1.2.0, plus the quality layer around it.
What's in 1.2.0
refineConfidencescoring for generic detectors, newflare-redact/mlsubpath,--refine-confidenceCLI flag. Verified in review: deterministic seeded training reproduces the shipped weights byte-for-byte; behavior probed on real vs benign strings (random token 0.99, git SHA 0.04, UUID 0.007).npm run benchmark:graphcovers what adapter users actually hit:redact()on flat strings and deep object graphs, and vault mint+restore round trips. Previously onlyscan()was measured.mapGraphtraversal (URL/URLSearchParamsrewriting,Errorprototypes with custom props, shared references, sparse arrays, symbol keys,Map/Set).Verification
npm pack --dry-runclean (103.4 kB, 114 files).Version bumped to 1.2.0 — merging publishes to npm via CI.